利用图像分析和深度学习预测油菜籽中总酚含量

Predicting total phenolic content in rapeseed using image analysis and deep learning

中文关键词:  油菜籽  图像分析  总酚含量  深度学习

英文关键词:rapeseed  image analysis  total phenolic content  deep learning

中文摘要:

      目的 建立了一种结合图像分析和深度学习的油菜籽中总酚含量的快速预测方法。方法 利用VGG19网络进行油菜籽图像籽粒特征的提取, 通过多个卷积层来学习油菜籽图像的特征, 并建立了回归模型用于预测油菜籽的总酚含量。共收集了100种油菜籽样本, 将油菜籽样本按照3:1的比例划分为训练集和测试集, 利用均方损失函数(MSELoss)和决定系数(r2)评估模型预测准确性。结果 测试集MSELoss=0.0085、r2=0.9914, 表明该预测模型具有一定的准确性和实用性。结论 本研究提出了一种快速、准确的评估油菜籽总酚含量的方法, 为油菜籽的总酚测定提供一种快速、准确的智能化检测方法。

英文摘要:

      Objective To establish a rapid method for predicting total phenol content in rapeseed by combining image analysis and deep learning. Methods The VGG19 network was used to extract features of rapeseed images, multiple convolutional layers were used to learn the features of rapeseed images, and a regression model was established to predict the total phenolic content of rapeseed. A total of 100 rapeseed samples were collected, and the rapeseed samples were divided into training sets and test sets at a ratio of 3:1. The mean square loss function (MSELoss) and coefficient of determination (r2) were used to evaluate the model prediction accuracy. Results On the test set, MSELoss was 0.0085, r2 was 0.9914, indicating that the prediction model had certain accuracy and practicality. Conclusion This study proposes a rapid and accurate method to evaluate the total phenolic content of rapeseed, which can provide a rapid and accurate intelligent detection method for the determination of total phenol of rapeseed.